Wednesday, March 26, 2014

Why High-Speed Rail Isn't Less Cost-Effective Than Other Transit Investments (Part One)


UCLA's Institute of Transportation Studies released a working paper report* two days ago, which makes the claim that local transit and active transportation upgrades are a more cost-effective way to reduce greenhouse gas emissions (GHGs) than California's planned high-speed rail (HSR) line from Los Angeles to San Francisco. And if their numbers are to be trusted, those local projects are way more cost-effective: when all is said and done, they calculate that HSR will cost approximately $361 per metric tonne of GHG reduction — 30 times more than a tonne sells for in California's cap-and-trade market — while Los Angeles' Orange Line Bicycle Path will actually save the state over $3,600 for every metric tonne of GHG reduction that they achieve. The Orange Line Busway and Gold Line light rail services both fall somewhere in between, but look much better than high-speed rail by this metric.

I'm going to take a look at why this is completely wrong in two parts. Today will be the (relatively) short post, regarding their misleading calculation of "Net User Costs" for high-speed rail and other transportation infrastructure.

Here's an adaptation of the table from their paper, which shows the costs of reducing greenhouse gas emissions. Notice that in the first column the full public cost is actually lowest for high-rail; only after they account for "Net User Costs" are factored in does HSR end up looking so bad.

Cost per metric tonne of GHG reduction:

Full Public Cost Full Public Cost Less Net User Costs
California High­-Speed Rail  $461 $361
Orange Line Busway $1,074 -$676
Gold Line Light Rail Transit $3,458 -$1,233
Orange Line Bicycle Path $1,061 -$3,670

So what are "Net User Costs"? They're the amount you save when you switch from one mode of travel to a cheaper mode. When I switch from driving to work to taking the Gold Line, I go from paying whatever it costs to operate my car to paying $1.50 for a one-way fare on LA's Metro system. When I switch from flying to taking high speed rail, my ticket cost drops from $97 to $81. It's in this conversion that UCLA's working paper so significantly misrepresents the private savings of local transit and bicycling relative to high-speed rail.

Acela high-speed rail.
There are a few problems here. The first is a bit abstract — it's actually the use of this metric in the first place. It completely ignores those that switch from driving to high-speed rail, because this is switching from a less expensive mode to a more expensive one. From the perspective of this paper, those diversions don't exist, or at best they're implicitly judged as mistakes. But people generally act pretty rationally, and someone that decides to trade in a car trip for a train trip is probably doing so for completely valid reasons. Maybe they don't want to deal with traffic, or they put a premium on their time, or they want to work during their trip. If the benefits of taking the train outweighed the costs, they'd either find another way to get there or they wouldn't make the trip at all. Reducing all decisions to the immediate monetary cost isn't very instructive, and the reality of millions of people choosing to use the Acela high-speed rail service in the Northeast, rather than driving, highlights that.

The second point is easier to understand, and probably more significant. Net user costs are an easy calculation for air to rail: the ticket cost of one minus the ticket cost of the other. Things get tricky when measuring the cost of switching from car to bike or bus, however. To calculate the savings, the report's authors need to know the cost of the new trip — easy: bikes are free and buses are $1.50 — and the cost of the car trip it replaced. The driving cost is found by multiplying the trip distance by the standard IRS mile rate, which is mostly used for deducting business-related transportation expenses. As of 2012, that rate was 55.5 cents per mile.

The reason this number is so inappropriate is that the mileage rate is "based on an annual study of the fixed and variable costs of operating an automobile," and the vast majority of people switching from their car to the train aren't getting rid of their cars (sadly), they're just using them less. Most of those "fixed and variable costs" — insurance and car payments in particular, as well as depreciation — are accruing no matter how much or how little you drive; most car owners that switch to transit or bicycling are only saving on gas, and perhaps parking in very dense areas. I wrote about this car ownership sunk cost bias last year. Though it's dependent on the driver, gas probably account for less than a quarter of most people's car expenses, so this study probably overstated the Net User Cost savings by a factor of 4, at least.

As a side note, since it's (falsely) assumed that no one is switching from driving to using HSR for their long-distance travel, the rail line doesn't benefit from this nifty accounting trick. That's convenient, since if driving actually did cost 55.5 cents per mile, it would cost $211 to drive from LA to SF — more than double a high-speed rail fare.

As someone who's been car-free for going on six years, I'm the last person to criticize someone trying to raise the profile of local transit and active transportation investments — I rely on them every day and frequently write about the need for more. Even though most people don't get rid of their cars when new transit services arrive, some do, and that's amazing. We should celebrate and encourage that. But at the same time, pitting different forms of clean, efficient transit against one another isn't productive, especially when those transit types serve entirely different purposes. I feel that this recent UCLA report understated the benefits of HSR while overselling the benefits of rail, bus, and bike infrastructure. In truth, they're both outstanding investments and perfect complements, and we should be striving to find ways to build more of each.

The second part to this critique will come soon, with a focus on the differences between high-speed rail and local transit infrastructure, and why they shouldn't be considered competitive with one another. 

*Streetsblog LA has a great summary of the report's findings.

9 comments:

  1. It doesn't do anyone any good to end up with pablum like, "In truth, they're both outstanding investments and perfect complements, and we should be striving to find ways to build more of each". The reality is that there are going to be X dollars generated in cap and trade dollars and that number will be arrived at completely independently of how many worthwhile uses for those funds exist. So, asking about the relative cost-effectiveness of each investments is a worthwhile exercise.

    ReplyDelete
    Replies
    1. Fair point. I'm not trying to argue that cost-effectiveness isn't relevant, or that there's not political decisions that need to be made. It is a bit of pablum, but that's in part because I intend to make my point a little more clear in the Part 2 post.

      First I just wanted to establish that the cost-effectiveness was not as different as it seems, at least by the metrics that the UCLA authors chose to use. As a bit of a preview of the next post, I plan to address your concern more directly, focusing on the potential for the CA HSR project to turn an operating profit. The UCLA authors assumed there would be zero operating profit, which further underestimates the cost-effectiveness of the project, but more importantly (in my opinion) is the fact that building HSR quickly means getting to the point where it can start earning a profit sooner. That, in turn, means a steady source of income that can fund local transportation projects.

      Relatedly, my view is that HSR should be viewed as directly in competition with highways and airports, not local transit, and should therefore be competing for dollars with THOSE types of projects. There's a lot of money that will be spent over the next 45 years on highways and airports, especially when you include maintenance, and any HSR expenditures should be viewed as a way to prevent more spending on highways and airports.

      It's true that there are limited cap-and-trade funds, but that's not the only place our transportation money comes from. To some degree it should all be looked at holistically, and when we consider spending some of the cap-and-trade funds on HSR we should look at the broader picture to figure out what that means for transportation expenditures as a whole. My feeling is that cap-and-trade ultimately frees dollars for other projects, and if it can run at an operational profit it should go forward.

      Delete
    2. Shane,

      We do consider those shifting from driving to high speed rail.
      We assume an automobile diversion rate of 80.7%, which we back out from the CAHSRA's figures. This means that the bulk of the ridership comes from avoided automobile trips: 27,481,366 for full build phase 1 in 2040 and 30,364,039 in 2060.

      We use the Authority's own estimate of avoided VMT for these trips, which we backed out as 149.76 avoided auto miles per HSR trip from the Authority's claim that 4,115,684,637 automobile miles would be avoided due to HSR and dividing that by the 27,481,366 trips. I was a bit surprised by this number (I thought it would be higher), but it passed a sanity check if many of the HSR trips are regional rather than end-to-end. If the CAHSRA has a better figure to use, I'll plug it into the model.

      As you might see, 149.76*0.555 = $83.1168, meaning only pennies of savings per HSR user (we use the $83 average ticket cost from the business plan). This is why we did the sensitivity analysis to the distance of the auto trip that HSR displaced, because it is a source of uncertainty.

      The CAHSRA used $0.24 per mile, and we mention the result using that per-mile cost in our paper.

      Delete
    3. Thanks for the response Juan -- I hope you don't take the criticisms personally, I just don't feel that HSR got a completely fair treatment in this analysis, whether it was intended or not.

      I think the analysis of car costs in general needs some heavy revision, because people saving 55 cents per mile by switching their trips from car to rail, bus, bike, or anything else just doesn't pass the sanity check, as you say. I would argue that you can only fairly use that number for those that actually get rid of their cars, not for those that replace trips. For those that just replace a trip or two a day, I suspect the actual personal savings is somewhere around 10-15 cents per mile -- this applies to both HSR and local transportation investments, of course.

      I hope you'll also take another look at the public cost, which is what I intend to write about next. You write that there is effectively no operational profit for HSR, but I think the 2014 Draft Business Plan very clearly refutes that idea. By my calculation, the cost of HSR ranges from as high as $46 billion to 2060 in the low ridership scenario (2013 dollars), to as low as $23 billion in the high ridership scenario. In a very generous view, in which high ridership goals are met and ancillary income accounts for 30% of total incomes (which is what the business plan says is about the highest percent possible), the total cost to 2060 could be as low as $9 billion. I think that's unrealistically optimistic, but it's worth mentioning. Especially in light of how much transit agencies seem to undershoot ridership projections these days.

      Delete
    4. I don't think I know an academic who takes constructive criticism about a paper personally. It's the nature of the business.

      Exhibit 6.6 of 2014 Draft Business Plan puts net operating profit at $165M
      http://www.hsr.ca.gov/docs/about/business_plans/FINAL_Draft_2014_Business_Plan.pdf. Incorporating this into the analysis brings the cost per ton down to $360. This seems low to me. Do you have other figures?

      Assuming a total project cost of $54,894,000,000 (exhibit 3.4 of the 2014 plan, which wasn't out when we started the analysis in January) rather than $68,000,000,000 (2012 Plan) yields a cost per ton of GHG reduced of $270. Do you have alternate figures for the cost of construction that aren't published by the CAHSRA?

      The presence/absence of urban transit facilities likely have more weight on an individual's decision to own & maintain a car or a household's decision on the ratio of cars to drivers than the HSR. I would argue that the $0.555 per mile cost is most applicable to the urban transit case, but we use it rather than the $0.24 per mile for HSR to be consistent.

      Delete
    5. I'll detail this more fully in my post, but check the following in the 2014 Draft Business Plan:

      Page 43 -- Revenue. Multiply each 5-year revenue number (except 2060) by 5 and sum them all together. Or for a slightly more precise number, average adjacent 5-year revenue numbers then multiply by 5. Doing things the latter way, I get revenue of $44 to $74 billion depending on ridership scenario. Multiply those revenues by 1.1 to account for a conservative amount of ancillary income, which brings range up to $48 to $81 billion.

      Page 49 -- Operations/Maintenance. Do the same thing. I get a range of costs from $27 to $34 billion.

      Page 37 -- Lifecycle costs (rolling stock). Estimates range from $6 to $8 billion.

      Subtract costs, including capital of $54.9 billion, from revenues, and you get a cost of between $15 and $41 billion through 2060. The medium ridership scenario costs about $29 billion.

      Delete
    6. And I agree on your assessment of urban transit having a greater impact on car ownership, so I'd have no problem with the mileage conversion rate being higher for urban transit than HSR, but I think even 24 cents per mile is high. How many people that use the Orange Line busway or bike lane actually get rid of the cars? I'd guess somewhere between 5 and 20 percent, but don't know offhand how you'd determine that, at least not with a lot of data collection. If you wanted to be consistent I think it's more realistic to use the 24 cents/mile for both HSR and urban transit, or even something like 28-30 cents for urban and 24 cents for HSR. I realize you're constrained by using the data provided and can't just make things up, but I think 55 cents just doesn't stand up to the slightest scrutiny, unless your average gas consumption is something like 10 mpg.

      Delete
    7. I would add that if you're looking at GHG emissions, you have to look at the *projected* GHG efficiency of the *future* automobile fleet. Relying on previous studies doesn't negate the need to do that.

      Frankly, the main result I get from reading the working paper is that the conclusions are extremely sensitive to a large number of different variables, *none of which you have good data for*.

      So, Juan, if you were being honest in your conclusions, you'd say "It is impossible with current data and analysis techniques to determine which of these is more GHG-efficient." I think there's probably an academic bias against that sort of conclusion to a paper, but when I read through the data and analysis, it starts to really jump out at me.

      The comparative results (what is better for GHG reduction) flip around completely depending on tiny changes in starting assumptions, and there's no valid way to calibrate those assumptions.

      Part of honesty in science is admitting when you *don't* know.

      Delete
    8. Juan,

      The calculation you explain seems suspect to me, because you're plugging in an $81 ticket cost for a trip that's only 150 miles long. This is 40-60% higher than the average revenue per unit of distance on the Shinkansen. It is about two and half times as high as intercity rail fare in France, Spain, and Germany; I do not have separate HSR and legacy intercity rail fare data, but in those countries a large majority of intercity travel is HSR. Follow links in paragraph 2 of this post of mine.

      It's unlikely California HSR will charge such a high ticket price. Indeed, the business plan still clings to an LA-SF estimate of about 100-odd dollars, for the full 400+ mile trip!

      More likely, the $81 average ticket includes all trips, including trips diverted from air. The trips diverted from highways are disproportionately shorter ones, such as LA-Fresno; the longer trips today, such as LA-SF, have significant air travel, so the average HSR trip could be much longer than the average HSR trip diverted from cars. There may also be a first class vs. coach issue: my guess is that first-class rail riders are disproportionately longer-distance business travelers whose alternative is flying rather than driving, and since average fares (at least the ones I have access to) include both travel classes, this again pushes down the fare for rail riders diverted from highways.

      Delete